The Method Of Estimating The Number Of Information Sources Based On CEEMDAN

Direction-Of-Arrial(DOA)estimation has become one of the research focuses of array signal processing,and it has a wide range of applications in radar detection,communication systems,aerospace,anti-jamming and other fields.In DOA estimation,the unknown number of sources may cause serious errors in classical DOA estimation algorithms.Therefore,the accurate estimation of the number of sources has become the key premise of DOA estimation.At present,many source number estimation algorithms are implemented under the condition that the number of sampling snapshots is large,but in practical engineering,the number of sampling snapshots is usually not large.The estimation performance of these algorithms in small snapshots is greatly reduced,and through analysis,it is found that when the number of sources is close to the number of elements used,the performance of many classical algorithms drops sharply or even fails.Aiming at the problem that the modal aliasing problem in the EMD process under small snapshot numbers leads to a high error rate of the decomposed IMF components and seriously affects the estimation performance of the number of sources,a complete integrated empirical mode based on adaptive white noise is proposed.Decomposition(Complete Ensemble Empirical Mode Decomposition with Adaptive Noise,CEEMDAN)source number estimation algorithm.The main contents of the paper are as follows:1.By consulting a large number of existing documents,a brief introduction is made to the research background of the subject and the current state of the research on the estimation of the number of sources.2.The theoretical basis of the array signal is summarized,including uniform linear array,uniform circular array and additive noise model.The principle of MUSIC algorithm is also introduced,and the key influencing factors of source number estimation are discussed.This paper focuses on the estimation algorithms based on information theory,the information theory criterion method based on diagonal loading,and the Gale circle method.The estimation performance of the number of sources with different numbers and background noises is compared and analyzed.3.Construct an improved HHT,namely: CEEMDAN+HT,to suppress the modal aliasing problem by introducing adaptive white noise in the EMD process.Firstly,CEEMDAN is used instead of EMD to obtain a series of IMF components,and then the false components are eliminated according to the method based on the correlation coefficient,and then the effective features of the instantaneous phase are extracted by HT,and finally the number of sources is estimated by BP neural network.4.Aiming at the problem that the modal aliasing problem in the EMD process under small snapshot numbers leads to a high error rate of the decomposed IMF components,which seriously affects the estimation performance of the number of sources,a method for estimating the number of sources based on CEEMDAN is proposed.Firstly,the overall process of the proposed modeling method is introduced in detail.The instantaneous phase of the signal is obtained by performing CEEMDAN on the array signal and taking the first3-order IMF components and then performing HT transform,and then extracting the feature vector of the instantaneous phase.Finally,the feature vector is substituted into the BP neural network for training,and the mathematical model obtained from the training can estimate the number of sources under small snapshots.Then,by comparing the EMD and EEMD algorithms to verify the superiority of CEEMDAN in small snapshots,and finally using theoretical data and RF anechoic chamber data to simulate the algorithm in this paper and the other five algorithms,and verify the proposed algorithm in small snapshots.The number and the number of sources are close to the number of array elements,especially when it is one less than the number of array elements,it has a good estimation performance.